Relativistic Semistochastic Heat-Bath Configuration Interaction

Abstract

In this work we present the extension of semistochastic heat-bath configuration interaction (SHCI) to work with any two-component and four-component Hamiltonian. The vertical detachment energy (VDE) of AuH2 – and zero-field splitting (ZFS) of NpO2 2+ is calculated by correlating more than 100 spinors in both cases. This work demonstrates the capability of SHCI to treat problems where both relativistic effect and electron correlation are important.

Consistent Second-Order Treatment of Spin–Orbit Coupling and Dynamic Correlation in Quasidegenerate N-Electron Valence Perturbation Theory

“… a Results are compared to the reference data from the SO-SHCI calculations . For all methods, the uncontracted ANO-RCC-VTZP and cc-pVTZ basis sets were used for the Np and O atoms, respectively. …”

“…Next, we use spin−orbit QDNEVPT2 to compute the energies of excited states originating from the zero-field splitting in the 2 Φ and 2 Δ terms of NpO 2 2+ , which exhibit strong electron correlation and spin−orbit coupling effects (Table 7). In this study, we benchmark against the recently published results of SO-SHCI calculations 57 that utilized the variational two-component treatment of relativistic effects with the DKH1 Hamiltonian. We note that the SO-SHCI calculations were performed in the (13e, 60o) active space, while our spin−orbit QDNEVPT2 methods correlated all 107 electrons in 433 molecular orbitals thus providing a more complete description of dynamic correlation.…”

“…Results are compared to the reference data from the SO-SHCI calculations. 57 For all methods, the uncontracted ANO-RCC-VTZP and cc-pVTZ basis sets were used for the Np and O atoms, respectively. In Table 8, we also report the excited-state energies for UO 2 + , which has the same electronic states and configuration as NpO 2 2+ .…”

“…The two-component relativistic theories can be broadly divided into two categories: (i) variational methods that incorporate relativistic effects in the self-consistent field (SCF) reference wave function − and (ii) perturbative approaches that introduce spin–orbit coupling as a posteriori correction together with dynamic correlation following a nonrelativistic SCF calculation. ,− The variational two-component methods can accurately describe electron correlation and spin–orbit coupling in molecules with elements from the entire periodic system, but their computational cost remains considerably higher than that of nonrelativistic theories. On the other hand, the perturbative methods have much lower computational cost, similar to that of nonrelativistic methods, but may be unreliable for the compounds with heavier elements where the relativistic effects become particularly strong.…”

Toward Accurate Spin–Orbit Splittings from Relativistic Multireference Electronic Structure Theory

“…As such, considerable work has been devoted to developing mean-field methods. On the four-component side, the Dirac–Hartree–Fock (DHF, or 4c-HF) and complete active space self-consistent field (4c-CASSCF) − can be routinely carried out in several packages. − Depending on whether a 2 c /4c-HF or CASSCF reference is used, these methods are further grouped into single- , or multireference approaches. − Multireference relativistic methods have received more attention than their nonrelativistic cousins, as many systems where a relativistic study is warranted, such as late-row transition metal, lanthanide, and actinide complexes, exhibit strong configuration mixing in the ground state, and as a result require the zeroth order wave function to be multiconfigurational. Examples of relativistic multireference theories include Fock-space multireference coupled cluster (FSMRCC), 4c internally contracted MRCI (ic-MRCI), CASPT2, NEVPT2, generalized van Vleck PT2 (GVVPT2), multireference Møller–Plesset (MRMP). , In several ways, these methods reflect one or more limitations of their nonrelativistic counterparts: (1) lack of proper scaling with system size, (2) difficulties with scaling to large active spaces, (3) numerical divergences arising from the intruder state problem, and (4) insufficient accuracy in the description of dynamical electron correlation.…”